Chapter 2 discussed the complexity of measuring VC Risk. VC Risk is a many-faceted concept, which is difficult to encapsulate in a single number. To provide a quantitative overview of VC Risk in the Asia-Pacific, we therefore proceed by identifying major categories of VC Risk and building the analysis up from that point. The approach adopted here focuses on achieving three goals for the resulting indicators: comprehensiveness;
transparency; and ease of replication. All three objectives are important for indicators that are to be both policy-relevant and potentially useful to value chain researchers inside and outside the Asia-Pacific region. A set of indicators that satisfies all three criteria will be easily interpretable for policymakers. Moreover, such a set of indicators will facilitate future work by researchers and policy experts, as well as timely and cost-effective updates of this Report if member economies consider that to be a fruitful avenue for future work.
It is important to highlight that measuring VC Risk needs to focus on the concept of risk, namely the possibility that an unforeseeable event occurs, and imposes costs on economic agents involved in value chain processes. VC Risk is therefore not just about the quality of infrastructure and other factors that influence the way in which value chains do business.
Many of those factors are more relevant to an assessment of resilience or strength, which will be addressed in Phase 2 of this Project. Moreover, an economy’s ability to respond to risks through a coordinated approach involving both public and private sector actors will also be examined in Phase 2. This first phase of the Project involves the analysis of risk in a pure sense, and not an economy’s ability to respond to it.
A NOTE ON PRINCIPAL COMPONENT ANALYSIS
The focus on risk in Phase 1 makes it possible to identify a relatively small number of indicators that capture the main dimensions of uncertainty that affect value chain processes.
To reduce the number of indicators to a manageable size while retaining a high degree of explanatory power, we have used a statistical technique known as Principal Component Analysis (PCA) to identify the indicators with the greatest explanatory power. .
PCA is a standard statistical technique that is widely used in economic analysis. It uses a mathematical procedure to produce an optimal summary indicator from a set of original indicators. The summary measure, known as the first principal component, is optimal in the sense that it accounts for the maximum possible variation in the original set of indicators.
No linear combination (such as an alternative weighted average) can account for a greater percentage of the variation in the original indicators than the first principal component. This technique has previously been used in the APEC context to measure multimodal transport connectivity (PSU, 2010) and by other international organizations such as the World Bank (Logistics Performance Index), and the United Nations (UNCTAD’s Liner Shipping Connectivity Index).
In the initial stage of the VC Risk project, a large set of potential indicators were analysed to help us better understand the nature and potential effects a wide range of scenarios could have on the value chain. As the analysis progressed, we observed a high degree of correlation within that indicator set, meaning that multiple data series were explaining the same effect. By using PCA, we were able to isolate a smaller subset of indicators that explained a majority of the risk in a clear, straightforward manner while removing those selections that added little to the overall analysis. Our results retain the same relative strength of explanation without the clutter of other unnecessary indicators and noise, making for an analysis that is easily understood yet robust.
Chapter 3: Approach to Measuring Value Chain Risk 11
The remainder of this Chapter discusses each uncertainty in turn, before presenting a methodology for rescaling and aggregating the individual data points into indices that can be used to assess VC Risk from a holistic perspective. Full details of data and sources are provided in the Appendix.
CATEGORIES OF VALUE CHAIN RISK
The literature review in Chapter 2 showed that there is no standard categorisation of VC Risks. We have therefore adopted a selective approach based on the previous literature. We focus on extracting risk categories that are important to business, as indicated in Chapter 2, but also relevant to policymakers seeking to understand and manage VC Risk at the economy level. To provide maximum transparency, we have tried to keep the number of categories to a manageable number. This approach facilitates accessibility from an end-user perspective, and makes replication and updating easier over the medium term.
Defining clearly what VC Risks are is difficult, as people will have different interpretations of what constitutes risk. Nevertheless, the following definitions serve as a useful starting point:
1. Natural Disaster Risks: the possibility that economic activity may be impeded by natural disaster.
2. Logistics and Infrastructure Risks: the set of disruptions that can occur to supply chain processes when the markets or actors that connect supply chain operators to each other do not perform as expected.
3. Market Risks: economic fluctuations that disrupt prices, output, or other economic fundamentals.
4. Regulatory and Policy Risks: unexpected changes in regulatory stance, or inconsistency in enforcement, can increase business uncertainty, and thus the transaction costs
associated with value chain processes.
5. Political Risks: the possibility that economic activity may be impeded by the occurrence of political or violent conflicts inside or outside the economy.
Natural Disaster Risks
Disasters and other natural phenomena have the potential to disrupt supply chains, because they can make it more difficult to move goods across borders or, in extreme circumstances, they may even shut down production entirely in important centres. The effect of floods in Thailand on value chains that use hard disks is a relatively recent example. Of course, many economies have shown over time that they have well-developed capacities to respond to natural disaster risks. The policy and private sector development factors that contribute to that capacity will be examined in Phase 2 of this Project, which deals with Value Chain Strength.
At this stage, the focus is exclusively on the underlying level of risk.
Measuring natural disaster risks is difficult, because a wide range of factors come into play.
One approach would be to include as many of those factors as possible in the indicator set, with the aim of achieving comprehensive coverage. However, this approach necessarily introduces complexity into the exercise. We prefer to focus on a small number of proxy measures, which is appropriate because data on natural disasters are often strongly correlated.
Moreover, not all natural disasters are common in the Asia-Pacific region, which means that
12 Quantitative Analysis on Value Chain Risks in the APEC Region
there is sometimes limited variation in the indicators across economies. The informational content of indicators that do not vary much is limited, and they can safely be excluded.
As a starting point, therefore, we consider a wide range of risks from the Project Concept Note, which we proceed to narrow down to a smaller number. We take eight into account:
earthquakes; floods; storms; mass geophysical movements; volcanos; wildfires; droughts; and extreme temperatures. PCA of the eight indicators shows that just three are responsible for a large proportion of the variation in the data: floods; storms; and earthquakes. Intuitively, these three are also the most important and commonly occurring natural disaster risks from a value chain point of view. Indeed, over 75% of the observed variation in the first principal component of the eight indicators is accounted for by a simple average of these three indicators.
To compose our measure of natural disaster risks, we therefore take a simple average of the following three indicators over the 20 year period 1992-2012 (after rescaling):
Total number of people affected by floods per year and per 100,000 population.
Total number of people affected by storms per year and per 100,000 population.
Total number of people affected by earthquakes per year and per 100,000 population.
The scales used for these measures—the number of people affected—are common in the literature on assessing the impacts of natural disasters. However, the same disaster would be regarded more serious if it occurred in an urban area than a rural one, because of increased population density. These measures are therefore imperfect, in the sense that they capture population effects, but not necessarily effects on transport infrastructure and connectivity, which are arguably more relevant to the operation of value chains. However, restrictions on the availability of comparable data across economies mean that these are the best data available to measure the importance of natural disaster risks in the value chain context at the present time.
Logistics and Infrastructure Risks
We use the category of logistics and infrastructure risks to refer to the set of disruptions that can occur to supply chain processes when the markets or actors that connect supply chain operators to each other do not perform as expected. Given the APEC context of this work, we focus on cross-border value chains. Logistics and infrastructure risks encompass the infrastructure risks category identified in the Project Concept Note, but also take a broader range of factors into account, including the performance of service providers.
Logistics is a broad concept, and one with many dimensions. The World Bank’s Logistics Performance Index (LPI), for example, considers six core dimensions of logistics performance. In the interests of simplicity and transparency, this report proposes a simpler approach using just two indicators from the LPI database. First, the LPI questionnaire asks respondents—who are professionals in the logistics industry—to evaluate the quality of trade and transport infrastructure in countries they do business with. This indicator therefore captures infrastructure risks as they relate to logistics in the cross-border context.2 Second,
2 The focus of this report is on value chains in the APEC context, so the cross-border element is important. This context is an important reason for preferring the LPI infrastructure quality measure to alternatives such as the World Economic Forum’s measures of infrastructure quality, which mostly deal with domestic structures.
Chapter 3: Approach to Measuring Value Chain Risk 13
the survey asks respondents to indicate the percentage of shipments that satisfy their firm’s quality criteria for delivery. Typically, logistics firms measure quality through such metrics as timeliness of delivery, and the state of the goods when they are delivered (e.g., undamaged, and not subject to criminal activity or loss). As a result, the inverse of the LPI measure—the percentage of shipments that do not meet firms’ quality criteria—can be considered to be a measure of the risk associated with logistics processes broadly conceived.
A higher score indicates a higher level of risk.
Importantly, measuring logistics and infrastructure risks in this way takes account of the full range of causes that can lead to disruptions in the logistics processes that make value chains work. For instance, an infrastructure failure typically results in delays and/or damage to goods, and so would tend to keep shipments from meeting firm quality criteria. Similarly, a deficiency in the market for transport or distribution services could lead to late delivery, which would again be recorded as a shipment that does not meet firm quality criteria. As a result of the comprehensive nature of these measures, and their inherent relationship to the idea of risk, it is unnecessary to include additional indicators—a wide range of factors are already accounted for by these indicators of logistics and infrastructure risks, and the addition of further data series will be of only limited informational benefit.
To compose our measure of logistics and infrastructure risks, we therefore use rescaled LPI measures of the quality of trade and transport infrastructure, and the percentage of shipments that do not meet firms’ quality criteria. In the APEC context, the shipment quality measure has also found recent application in the closely linked area of supply chain connectivity: it is one of the external indicators of building infrastructure and capacity presented in the mid-term review of the Supply Chain Connectivity Framework Action Plan (PSU, 2012).
The Project Concept Note identifies a range of possible data sources for its category of infrastructure risks: road and rail network density; the quality of road, rail, and port infrastructure; fixed and mobile telephone subscribers; the number of internet users; and access to improved water. In terms of the broader VCR project, these indicators are more relevant to the assessment of value chain strength rather than risk: they do not directly assess the likelihood of events with negative impacts. In any case, a simple average of the indicators proposed here accounts for over 50% of the observed variation in an optimal summary index of the full set of indicators, produced as in the previous section by PCA.
Market Risks
Market risks of various types have considerable potential to disrupt value chain processes.
The term is used broadly to apply to economic fluctuations that disrupt prices, output, or other economic fundamentals. For example, the global financial crisis and resulting collapse of trade seriously upset—albeit temporarily—cross-border value chain processes around the world, including in the Asia-Pacific. Economic crises, as well unforeseen fluctuations more generally, have the capacity to decrease trade and investment flows, with follow-on international effects to employment and income.
Although other types of infrastructure, such as access to electricity or water, can impede economic activity, they are not particular to cross-border value chains, and so are not considered here. Moreover, general infrastructure indicators have a closer correspondence with value chain strength (Phase Two of this project) rather than value chain risk.
14 Quantitative Analysis on Value Chain Risks in the APEC Region
As was the case for natural risks, economic fluctuations that can affect value chains come in many shapes and forms. One approach would be to include a wide range of indicators in an effort to be comprehensive. However, we have concluded that such a complex index would have relatively little added value compared with a simpler version in which just a few proxy indicators are selected which captures the most important economic fluctuations from a VC Risk perspective. Again, economic fluctuations are often correlated, so the informational content of an additional indicator can sometimes be relatively limited.
As an exploratory exercise, we have considered thirteen elements of market risk, based in part on the preliminary ideas presented in the Project Concept Note: fluctuation of the Consumer Price Index; instability of nominal GDP growth; instability of the nominal interest rate; instability of the share price index; cereal import dependency ratio; domestic food price level volatility index; proportion of oil imports to total imports; sovereign rating; central government gross debt as a proportion of GDP; foreign reserves in proportion to monthly imports; net international investment position as a percentage of GDP; instability of the current account balance as a proportion of GDP; and instability of domestic credit provided by the banking sector as a proportion of GDP.
PCA shows that most of the common variation in these indicators is accounted for by just three: fluctuation of the CPI; sovereign ratings; and the net international investment position.
A simple average of these three indicators accounts for over 75% of the observed variation in the first principal component of the thirteen series.
Our market risks index is therefore a simple average of the following three indicators (after rescaling):
Instability of the CPI (a 5-year simple average).
Sovereign ratings (an average of the 3 ratings components from Moody's, S&P and Fitch).
Net international investment position (the difference between an economy’s external financial assets and liabilities) as a percentage of GDP.
Regulatory Risks
Regulatory and policy risks can affect value chain performance as unexpected changes in regulatory stance which are not in accordance with international norms; which foster protectionist policies; or allow for inconsistent enforcement, can increase business uncertainty and thus the transaction costs associated with value chain processes. It is therefore appropriate to include information on such risks in the VCR project, even though they are difficult to quantify.
For the previous categories of VC Risks, we have used a limited range of proxy indicators to capture the essential aspects of the relevant risks, rather than adopting an approach in which a full range of correlated indicators is included. The case of regulatory risks is made much simpler than the others by the existence of the World Bank’s World Governance Indicators (WGIs). The WGIs cover six core elements of governance, each consisting of an index constructed from a large number of underlying data series using sophisticated statistical techniques. The result is a set of indices ranging from -2.5 to 2.5, with a higher score indicating better governance.
Chapter 3: Approach to Measuring Value Chain Risk 15
Two of the WGI indices are of particular interest for regulatory risks: rule of law; and control of corruption. Economies with a stronger rule of law are less likely to impose arbitrary and unforeseen regulatory changes, and those where corruption is limited provide firms with greater transparency and certainty when it comes to the costs linked with common transactions. To convert the indicators to measures of regulatory risk rather than regulatory good practice, each is multiplied by negative one: a higher score therefore implies a weaker rule of law (greater risk), or a greater prevalence of corruption (again, greater risk). These two indicators together capture the most important aspects of regulatory risk from the point of view of value chain processes.
Our regulatory risks indicator is therefore a simple average of the following indicators, rescaled as discussed below:
The WGI rule of law index.
The WGI control of corruption index.
To show that these indicators indeed capture a wide range of regulatory risks, we again conduct PCA using a broader set of indicators, as foreshadowed in the Project Concept Note:
the six World Governance Indicators (rule of law, regulatory quality, control of corruption, political stability and the absence of violence, voice and accountability, and government effectiveness); the number of free trade agreements and economic partnership agreements in force; and the numbers of international investment agreements and double taxation agreements concluded. The simple average of the two series we have proposed accounts for over 92% of the observed variation in the first principal component of that set of indicators.
We are therefore confident that using just these two indicators captures a wide range of regulatory risks that have the capacity to affect value chain performance.
Political Risks
Armed conflict, terrorism, and political instability can affect value chain processes by increasing the time and cost of transactions, and reducing reliability. It is therefore important to consider them as part of the VC Risk assessment process, even though they are sometimes difficult to quantify.
The types of political risks that can affect value chains are numerous and include all forms of
The types of political risks that can affect value chains are numerous and include all forms of